Signal definition and resource management techniques for upcoming communication and broadcasting systems

Texte intégral

(1)Signal definition and resource management techniques for upcoming communication and broadcasting systems Charbel Abdel Nour. To cite this version: Charbel Abdel Nour. Signal definition and resource management techniques for upcoming communication and broadcasting systems. Signal and Image processing. Université de Bretagne Sud, 2020. �tel-03052855v2�. HAL Id: tel-03052855 https://hal-imt-atlantique.archives-ouvertes.fr/tel-03052855v2 Submitted on 4 Jun 2021 (v2), last revised 2 Jul 2021 (v4). HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés..

(2) 1. Université de Bretagne-Sud Ecole Doctorale – MathSTIC. Signal definition and resource management techniques for upcoming communication and broadcasting systems. Habilitation à Diriger des Recherches HDR Presented by. Charbel Abdel Nour Prepared at. IMT Atlantique Department : Electronics Laboratory : Lab-STICC / IAS team. Defended online 30/11/2020. Defense committee : Gianluigi Liva Christophe Jego Carlos Bader Charly Poulliat Emmanuel Boutillon Catherine Douillard. Head of transmission group, DLR, Germany Professor at INP/ENSEIRB-MATMECA, Bordeaux Professor at ISEP, Paris Professor at INP/ENSEEIHT Toulouse Professor at UBS, Lorient Professor at IMT Atlantique, Brest. Reviewer Reviewer Reviewer President Examiner Examiner.

(3)

(4) Contents Contents. i. List of Figures. v. Acronyms. vii. Short introduction. xv. Curriculum Vitæ 1 Strong points/Summary . . . . . . . . . . . . . . . . . . 2 Education and career . . . . . . . . . . . . . . . . . . . . 2.1 Education . . . . . . . . . . . . . . . . . . . . . . 2.2 Career . . . . . . . . . . . . . . . . . . . . . . . . 3 Research activities . . . . . . . . . . . . . . . . . . . . . 4 Advisorship/Tutorship . . . . . . . . . . . . . . . . . . . 5 Contribution to research projects . . . . . . . . . . . . . 6 Collaborations . . . . . . . . . . . . . . . . . . . . . . . 7 Remarkable results . . . . . . . . . . . . . . . . . . . . . 8 Teaching activities . . . . . . . . . . . . . . . . . . . . . 8.1 Topics and courses . . . . . . . . . . . . . . . . . 8.2 Teaching and administrative responsabilities . . 8.3 Other responsabilities, activities and distinctions 9 Publications . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. 1 1 2 2 2 2 4 5 6 8 9 9 10 10 11. Introduction to my research activities. 13. 1 Algorithms for coded modulations with(out) MIMO 1.1 Introduction and motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Diversity, constellation and mapping for broadcasting standards . . . . . . . . 1.2.1 Context and prior art . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2.1.1 Coded modulation and diversity techniques . . . . . . . . . . 1.2.1.2 Signal space diversity and receiver implementation . . . . . . 1.2.1.3 Channel emulation/simulation . . . . . . . . . . . . . . . . . 1.2.1.4 Non-uniform constellations and corresponding demappers . . 1.2.1.5 Advanced interleaving for improved system performance . . . 1.2.2 Performed work and contributions . . . . . . . . . . . . . . . . . . . . 1.2.2.1 Signal space diversity design and action in DVB standardization. 15 15 16 16 16 17 17 18 19 19 19. i.

(5) ii. Contents. 1.3 1.4. 1.5. 1.6. 1.2.2.2 Signal space diversity and receiver implementation . . . . . . 1.2.2.3 Channel emulation/simulation . . . . . . . . . . . . . . . . . 1.2.2.4 Non-uniform constellations and corresponding demappers . . 1.2.2.5 Advanced interleaving for improved system performance . . . MIMO related studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 Context and prior art . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Performed work and contributions . . . . . . . . . . . . . . . . . . . . Improving binary Turbo codes . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.1 Context and prior art . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 TC design: Performed work and contributions . . . . . . . . . . . . . 1.4.3 Actions in the 3GPP consortium . . . . . . . . . . . . . . . . . . . . . 1.4.4 3GPP standardization discussions and conclusions . . . . . . . . . . . 1.4.5 Performed work and contributions in designing beyond 5G high throughput Turbo codes/decoders . . . . . . . . . . . . . . . . . . . . . . . . . Study of non-binary codes and non-binary coded modulations . . . . . . . . . 1.5.1 Context and prior art . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.1.1 Non-binary LDPC codes and coded modulation . . . . . . . 1.5.1.2 Non-binary LDPC codes and MIMO . . . . . . . . . . . . . . 1.5.1.3 Non-binary convolutional and Turbo codes . . . . . . . . . . 1.5.2 Performed work and contributions . . . . . . . . . . . . . . . . . . . . 1.5.2.1 Design of non-binary LDPC codes . . . . . . . . . . . . . . . 1.5.2.2 SSD and non-binary LDPC codes: Study of an association . 1.5.2.3 Non-binary LDPC codes and MIMO: Proposal of a joint framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.2.4 Design and proposal of non-binary convolutional and Turbo codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 27 28 29 29 41 41 41 47 47 48 62 63 65 81 81 81 82 82 83 83 83 83 90 92. 2 Algorithms for filtered waveforms 95 2.1 Introduction and motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 2.1.1 Post-OFDM waveforms: Improvement by filtering . . . . . . . . . . . 95 2.1.2 Post-OFDM waveforms: Faster than Nyquist signalling . . . . . . . . 97 2.2 Performed work and contributions . . . . . . . . . . . . . . . . . . . . . . . . 98 2.2.1 Design of filtered waveforms . . . . . . . . . . . . . . . . . . . . . . . . 98 2.2.1.1 Design and algorithmic contributions . . . . . . . . . . . . . 98 2.2.1.2 Architecture, implementation and system aspects . . . . . . 99 2.2.1.3 Implementation of FBMC-OQAM and OFDM transmitters . 116 2.2.1.4 Novel UF-OFDM transmitter . . . . . . . . . . . . . . . . . . 118 2.2.1.5 Complete demonstrator platform . . . . . . . . . . . . . . . . 133 2.2.2 Performed work and contributions on FTN signalling for FBMC/OQAM134 2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 3 Algorithms for NOMA systems 3.1 Introduction and motivation . . . . . . . . . . . . 3.1.1 Resource allocation . . . . . . . . . . . . . 3.1.1.1 Resource allocation for OMA . . 3.1.1.2 Resource allocation for NOMA . 3.2 Performed work and contributions for NOMA . . 3.2.1 Cognitive radio-like scenario . . . . . . . . 3.2.2 Classical communications system scenario 3.2.3 Extension to single-user MIMO . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. 145 145 146 146 147 148 148 149 154.

(6) Contents. iii. 3.2.4 3.2.5 3.2.6. 3.3. Allocation dedicated to the reduction of BS power consumption . . . . 154 Joint broadcast-broadband operation . . . . . . . . . . . . . . . . . . . 155 NOMA for distributed antenna systems . . . . . . . . . . . . . . . . . 155 3.2.6.1 New power minimization techniques for OMA and NOMA with power-constrained DAS . . . . . . . . . . . . . . . . . . 171 3.2.6.2 Resource allocation for mixed traffic types in DAS using NOMA172 3.2.6.3 Mutual SIC strategies in NOMA to enhance the spectral efficiency of coordinated multi-point systems . . . . . . . . . . . 189 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190. 4 Future works 191 4.1 Ongoing work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 4.1.1 Coding and modulation-related: Towards Tb/s turbo decoders . . . . 191 4.1.1.1 Motivation and introduction . . . . . . . . . . . . . . . . . . 191 4.1.1.2 Way forward and study items . . . . . . . . . . . . . . . . . . 192 4.1.2 FBMC-related: Towards a complete solution for waveform design beyond 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 4.1.2.1 PAPR reduction for FBMC with short ProFs . . . . . . . . . 194 4.1.2.2 Enrichment of the proof-of-concept waveform platform . . . 194 4.1.3 NOMA-related: Drone-assisted communications . . . . . . . . . . . . . 195 4.1.3.1 In-band full-duplex and backhaul-constrained drone-enabled networks using NOMA . . . . . . . . . . . . . . . . . . . . . 195 4.1.3.2 Drone placement for mutual SIC optimization in two-cell NOMA CoMP systems . . . . . . . . . . . . . . . . . . . . . . . . . . 196 4.2 Long term future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 4.2.1 Towards efficient Tb/s turbo decoders . . . . . . . . . . . . . . . . . . 197 4.2.2 FEC code construction using genetic and AI-based algorithms . . . . 199 4.2.3 Massive MIMO FBMC-OQAM . . . . . . . . . . . . . . . . . . . . . . 199 4.2.4 D2D communications underlaying cellular networks . . . . . . . . . . . 200 4.2.5 Support of grant-free or uncoordinated multiple access through learning algorithms and/or NOMA . . . . . . . . . . . . . . . . . . . . . . . . . 201 4.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 List of publications. 205. References. 217.

(7)

(8) List of Figures 1 3 4 5 6. Education and career at a glance . . . . . . . . . Summary of advisorship/tutorship activity . . . Collaboration map . . . . . . . . . . . . . . . . . Google scholar citations and h-index . . . . . . . Map of the research activities since the beginning. 1.1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . of my PhD. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. DVB Scene magazine cover from 28/02/2011 with emphasis on rotated constellations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 DVB-T2 transmitter, channel emulator and receiver of Teamcast integrating proposed BICM receiver demonstrated at IBC 2010. . . . . . . . . . . . . . . . . . . 1.3 The developed GUI for the demonstrator. . . . . . . . . . . . . . . . . . . . . . . 1.4 Performance comparison of the enhanced TC with the Multi-Edge LDPC code in AWGN channel for coding rates ranging from 1/5 to 8/9 in terms of BLock Error Rate vs Es/N0. QPSK modulation, block size K around 100 bits. . . . . . . . . 1.5 Performance comparison for the first transmission of the enhanced TC with the polar code in AWGN channel for coding rates ranging from 1/5 to 8/9 in terms of required Es/N0 to achieve 1% and 0.1% of BLER. QPSK modulation, block size K ranging from 32 to 1024 bits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Performance comparison for up to four retransmissions of the enhanced TC with the polar code in AWGN channel for coding rates ranging from 1/12 to 2/3 in terms of required Es/N0 to achieve 1% and 0.1% of BLER. QPSK modulation, block size K ranging from 32 to 1024 bits. . . . . . . . . . . . . . . . . . . . . . . 1.7 The joint Factor Graph representation of the detector and the decoder (example given for N T = N R = 4, D = 1, Hadamard STBC). . . . . . . . . . . . . . . . . 1.8 Comparison between the two template structures of a classical accumulator and the proposed one. The corresponding trellises are represented for q=4. . . . . . . 1.9 Performance comparison between the best state-of-the-art codeover Z(64), the best code over GF(64) and the proposed code – named C3 – over GF(64) from the application of the new design procedure. 64-QAM constellation, R = 1/2 over an AWGN channel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10 Frame error rate performance comparison with ARP interleaver over an AWGN channel with 64-QAM modulation, Ks = 160 GF(64) symbols and R = 1/3 for two convolutional codes, C1 and C3, in GF(64) with and without symbol transformation. 1.11 Frame error rate performance comparison of NB-TC defined over GF(64), Ks = 164 GF(64) symbols, and 5G-polar code, Kb = 984 bits with different coding rates, AWGN channel and 4-QAM constellation. . . . . . . . . . . . . . . . . . . v. 2 4 7 11 14 20 28 28. 62. 63. 64 84 90. 91. 92. 92.

(9) vi. 2.1. List of Figures. 2.6 2.7 2.8. PSD comparison of OFDM and FBMC/OQAM using a short and a long prototype filter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 OFDM transmitter and PPN implementation of the FBMC/OQAM transmitter. 96 Timing offset evaluation in terms of measured SIR for OFDM and FBMC with the considered short ProFs. The effect of different implementations is also evaluated. 99 Evaluation of the impact of NG (number of non-zero filter coefficients) on the SIR for different ProFs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 BER evaluation of OFDM and FBMC with different ProFs and implementations in presence of AWGN, for ETU static channel. . . . . . . . . . . . . . . . . . . . 100 Optimized FBMC/OQAM hardware architecture using pruned IFFT algorithm . 116 3-way Decomposition of the UF-OFDM symbol . . . . . . . . . . . . . . . . . . . 118 Demonstration setup with front-end interface. . . . . . . . . . . . . . . . . . . . . 133. 3.1 3.2 3.3 3.4. The principle of PD NOMA [182, 183, 184]. . . . . . . . . . . . The problem of resource allocation for of PD NOMA. . . . . . The different scenarios for resource allocation of of PD NOMA. DAS cell with two power-limited RRHs (RRH 1 and RRH 4). .. 4.1 4.2. System model for IBFD backhaul-constrained UAV assisted networks. . . . . . . Illustration of the two-cell JT system with the functional base station a1 , the saturated BS in cell 2, the UAV working as MBS a2 , and the three colored user regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . UXMAP architecture framework, prime parameters and their effects. . . . . . . . FD inband underlay communication sharing the uplink resource of a cellular user.. 2.2 2.3 2.4 2.5. 4.3 4.4. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. 146 147 149 171 196 197 198 201.

(10) Acronyms 2G 2nd Generation mobile networks 3GPP 3rd Generation Partnership Project 4G/LTE 4th Generation mobile networks / Long Term Evolution 5G 5th Generation mobile networks ACLR adjacent channel leakage ratio ACS Add-Compare-Select ADC Analog-to-Digital Converter AFB Analysis Filter Bank AI Artificial Intelligence API Application Programming Interface ARP Almost Regular Permutation AS Antenna Selection ASIC Application-Specific Integrated Circuit AWGN Additive White Gaussian Noise BBU BaseBand Unit BCH Bose Ray-Chaudhuri Hocquenghem BCJR Bahl Cocke Jelinek Raviv BE Best Effort BER Bit-Error Rate BF-OFDM Block-Filtered-Orthogonal Frequency-Division Multiplexing BICM Bit-Interleaved Coded Modulation BICM-ID Bit-Interleaved Coded Modulation with Iterative Demodulation vii.

(11) viii. BLER BLock-Error Rate BMI Bitwise Mutual Information BPSK Binary Phase Shift Keying BS Base Station C-RAN Cloud Radio Access Network CAS Centralized Antenna Systems CC Convolutional Codes CCDF Complementary Cumulative Distribution Function CFO Carrier Frequency Offset CFR Channel Frequency Response CIR Channel Impulse Response CM Complex Multiplier CoMP Coordinated MultiPoint CP Cyclic Prefix CPE Common Phase Error CSD Canonical Signed Digit CSI Channel State Information D2D Device-to-Device DA Deferred Acceptance DAC Digital-to-Analog Converter DAS Distributed Antenna Systems DCT Discrete Cosine Transform DDR Double Data Rate DFE Decision Feedback Equalizer DIF Decimation In Frequency DMA Direct Memory Access DMSIC Double Mutual Successive Interference Cancellation DMT Diversity-Multiplexing Tradeoff DSP Digital Signal Processor DRP Dithered Relative Prime DRS Demodulation Reference Signal. Acronyms.

(12) Acronyms. DVB Digital Video Broadcasting DVB-NGH Digital Video Broadcasting New Generation Handheld DVB-RCS2 Digital Video Broadcasting Return Channel Satellite 2nd generation DVB-T2 Digital Video Broadcasting Terrestrial 2nd generation EIC Equalization with Interference Cancellation eMBB enhanced Mobile BroadBand EMCM Even Multiplierless Constant Multiplication EPA Extended Pedestrian A ETU Extended Typical Urban EVA Extended Vehicular A EVM Error Vector Magnitude EXIT EXtrinsic Information Transfer F-OFDM Filtered-Orthogonal Frequency-Division Multiplexing FB Filter-Bank FC-OFDM Flexible-Configuration Orthogonal Frequency-Division Multiplexing FBMC Filter Bank Multi-Carrier FD Full-Duplex FDA Frequency Domain Approximation FDC Frequency Domain Compensation FEC Forward Error Correction FFT Fast Fourier Transform FIFO First-In First-Out FIR Finite Impulse Response FPA Fixed Power Allocation FPGA Field-Programmable Gate Array FS Frequency Spread FSPA Full Search Power Allocation FTN FasterThan-Nyquist FTPA Fractional Transmit Power Allocation GUI Graphical User interface. ix.

(13) x. GFDM Generalized Frequency Division Multiplexing HARQ Hybrid Automatic Repeat reQuest HFS Half Frequency Shift HPA High Power Amplifier IAI Inter-Antenna-Interference IAM Interference Approximation Method IBC International Broadcasting Convention IBFD In-band Backhaul Full-Duplex ICeI Inter-Cell Interference ICI Inter-Carrier Interference IFFT Inverse Fast Fourier Transform IIR Infinite Impulse Response IMT2020 International Mobile Telecommunications-2020 IoT Internet-of-Things IOTA4 Isotropic Orthogonal Transform Algorithm 4 IR Incremental Redundancy ISI Inter-Symbol Interference IUI Inter-User Interference JFG Joint Factor Graph JT Joint Transmission LDM Layer Division Multiplexing LDPC Low density Parity Check LO Local Oscillator LoS Line-of-Sight LPR Linear Phase Rotation LTE Long Term Evolution LUT Look-Up Table M2M Machine-to-Machine MAB Multi-Armed Bandits MAP Maximum A Posteriori. Acronyms.

(14) Acronyms. Max-Log-MAP Maximum A Posteriori with Max-Log approximation MBB Mobile BroadBand MBS Master Base Station MCC Mission Critical Communication MCM Multiplier-less Constant Multiplication MI Mutual Information MIMO Multiple-Input Multiple-Ouput MISO Multiple-Input Single-Ouput ML Maximum Likelihood MMB4 Martin–Mirabassi–Bellanger 4 MMC Massive Machine Communication mMIMO massive Multiple-Input Multiple-Output MMSE Minimum Mean Square Error MRC Maximum Ratio Combining MTC Machine Type Communications MTD Machine-Type-Devices MUST MultiUser Superposition Transmission NB Non-Binary NB-CC Non-Binary Convolutional Codes NLOS Non-Line-of-Sight NMSE Normalized Mean Square Error NN Neural Network NOMA Non Orthogonal Multiple Access NPR1 Near Perfect Reconstruction 1 NR New Radio NUC Non-Uniform Constellation OFDM Orthogonal Frequency-Division Multiplexing OMA Orthogonal Multiple Access OMCM Odd Multiplierless Constant Multiplication OOBPL Out-Of-Band Power Leakage OS Overlap-Save. xi.

(15) xii. Acronyms. OSB Overlap-Save-Block OQAM Offset Quadrature Amplitude Modulation P-OFDM Pulse-shaped Orthogonal Frequency-Division Multiplexing PAM Pulse-Amplitude Modulation PAPR Peak-to-Average Power Ratio PAS Probabilistic Amplitude Shaping PD PPower Domain PER Packet Error Rate PF Proportional Fairness PHYDYAS Physical Layer For Dynamic Spectrum Access And Cognitive Radio PPN PolyPhase Network ProF Prototype Filter PSD Power Spectral Density PTC Precoded Turbo Code PU Primary User QAM Quadrature Amplitude Modulation QMF1 Quadrature Mirror Filter 1 QoS Quality of Service QPP Quadratic Permutation Polynomial QPSK Quadrature phase-shift keying RA Real Addition RAM Random Access Memory RB Resource Block RF Radio Frequency RI Roll-off Interval RL Reinforcement Learning RM Real Multiplication RRH Remote Radio Head RT Real Time SBF Subband Prototype Filter.

(16) Acronyms. SCMA Sparse Code Multiple Access SC Successive Cancellation SC-OFDM Single-Carrier Orthogonal Frequency-Division Multiplexing SeI Self Interference SFN Single Frequency Network SI Side Information SIC Successive Interference Cancellation SINR Signal-to-interference-plus-noise ratio SISO Soft-Input Soft-Output SIR Signal-to-Interference Ratio SLM Selective Mapping SMT Staggered Modulated Multitone SNR Signal-to-Noise Ratio SP-RAM Single-Port Random-Access Memory SoC System on Chip SOVA Soft Output Viterbi Algorithm SQNR Signal-to-Quantization-Noise Ratio SSD Signal Space Diversity SU Secondary User STBC Space-Time Block-Coding TA Timing Advance TC Turbo Code TDD Time-Division Duplexing TDW Time Domain Windowing TFL1 Time Frequency Localization 1 TMSIC Triple Mutual Successive Interference Cancellation TTI Time Transmission Interval TUKL Technical University of Kaiserslautern UAV Unmanned Aerial Vehicle UE User Equipment UF-OFDM Universal-Filtered Orthogonal Frequency-Division Multiplexing. xiii.

(17) xiv. Acronyms. UFMC Universal Filtered Multi-Carrier UMTS Universal Mobile Telecommunications System URLLC Ultra-Reliable Low Latency Communications UXMAP Unrolled X Maximum A Posteriori V2I Vehicular-to-Infrastructure V2X Vehicular-to-Anything V2V Vehicular-to-Vehicular VHDL VHSIC (Very High Speed Integrated Circuit) Hardware Description Language ZF Zero-Forcing ZT-OFDM Zero-Tail Orthogonal Frequency-Division Multiplexing.

(18) Short introduction This document represents my Habilitation thesis manuscript. It summarizes my academic and professional career including recent and future research activities. It is organized as follows: - A first part includes my Curriculum Vitae; - A second short part gives a high level view on my research activities and articulates the rest of the document into addressed topics; - Chapter 1 presents the first topic related to the study and proposal of algorithms for coded modulations with(out) MIMO; - Chapter 2 presents the second topic related to the study and proposal of algorithms for filtered waveforms; - Chapter 3 presents the third topic related to the study and proposal of algorithms for non-orthogonal multiple access systems; - The final Chapter 4 provides a brief review of planned short and long-term future works.. xv.

(19)

(20) Curriculum Vitæ Current professional address: IMT Atlantique — Electronics department Lab-STICC UMR CNRS 6285 Technopôle Brest-Iroise — CS 83818 29238 Brest Cedex 3 France Phone : +33 2 29 00 15 94 Fax : +33 2 29 00 11 84 Email : Charbel.Abdelnour@imt-atlantique.fr Web :. 1. Strong points/Summary. •. 15+ years in R&D in the fields of digital communications and broadcasting.. •. Active participation (35+ technical contributions) to the technical groups of 3GPP and three different DVB standards.. •. Contribution to 18+ different R&D projects.. •. Participation to 5 different teaching modules in undergraduate engineering degree. The creation of new teaching courses including advanced ones for last year students.. •. In charge of the teaching module related to Electronics for the second year engineering students from 2014 till 2019. Since 2019, in charge of the digital and analog integrated circuits course.. •. Large experience in managing a R&D team and tutorship: 7 post-doctoral researchers, 14 PhD students (including 9 defended), 4 R&D engineers and several Master and undergraduate students.. •. 18 patent filings including 8 extended to multiple countries and 1 patent adopted and recognized as essential in 3 different DVB standards.. •. 24 journal publications, 2 book chapters, 5 invited talks and 65 publications in conferences including 2 articles attributed best paper awards. 1.

(21) 2. Curriculum Vitæ. •. 2. 5+ demonstrations in forums including the international Broadcasting Convention (IBC) in 2010 (250+ visitors) and at the Mobile World Congress (MWC) in 2015 (1000+ visitors including the EU commissioner in charge of the digital technology).. Education and career. Figure 1: Education and career at a glance. 2.1. Education. Oct. 1997Oct. 2002. Engineering diploma in electronics, computer and communications. Lebanese university, faculty of engineering branch II, Roumieh, Lebanon.. Oct. 2002Oct. 2003. Master Degree in electronics and image processing. Université de Valenciennes et du Haînaut Cambrésis (UVHC), Valenciennes, France.. Oct. 2003Jan. 2008. PhD in electronics and communications: “Iterative methods for improving error correcting performance of spectrally efficient coded transmissions”, Telecom Bretagne, Brest, France.. 2.2. Career. Jun. 2007Nov. 2011. Post-doctoral researcher in telecommunications. Telecom Bretagne, Brest, France.. Nov. 2011Present. Associate Professor in electronics and telecommunications. Telecom Bretagne/IMT Atlantique, Brest, France.. 3. Research activities. My research activities take place in the “Laboratoire des Sciences et Techniques de l’Information, Communication et Connaissance” (Lab-STICC) within the team specialized in the Interaction between Algorithm and Silicon (IAS) where algorithms are designed while taking into account their future implementation. They target the elaboration of new algorithms in the field of digital communications and broadcasting. In particular, the study of: •. Binary and non-binary error correcting codes (turbo, LDPC, polar codes etc.) and their associated decoding..

(22) 3. Research activities. 3. •. Digital modulations and Continuous Phase Modulation (CPM), etc.. •. Multi-antennas or Multiple Input Multiple Output (MIMO) systems.. •. Waveform design and multi-carrier modulations such as Filter-Bank MultiCarrier with Orthogonal Quadrature Amplitude Modulation (FBMC/OQAM), etc.. •. Diversity techniques (signal space, interleaving, space-time codes, etc.).. •. Resource and power allocation for (Non)-Orthogonal Multiple Access techniques ((N)OMA).. •. Joint design of several physical layer components; system optimization for standards (LTE, DVB-T2, etc.); simulation methods of correlated fading.. Figure 2: Guiding steps of research activity These algorithmic studies have the particularity of following partially or totally, depending on the study at hand, an IAS approach summarized by the steps depicted in Fig. 2. At the start of every study, there is an original idea that is often the fruit of an out-of-themainstream thinking. Deeper studies are needed hereafter to validate the originality of this idea and its positive impact on the obtained measurable results for the targeted application. If possible, implementation constraints would be already taken into consideration when elaborating corresponding algorithms following the IAS approach. In other words, choices would be made favouring hardware friendly functions. The work can be performed in the context of a collaborative/industrial project depending on the application. This allows providing the necessary funding to finance a Master degree student, a PhD student or a Post-doc fellow. Once validated via simulation results showing the superiority of the idea when compared to existing solutions in state of the art, the study can be the subject of a dissemination via one or more publications and potentially via patent filing. A second phase can be initiated in collaboration with IAS team members specialized in hardware implementation where at first, fully hardware compliant algorithms are devised. Then architectural templates are explored depending on the translated application requirements into hardware constraints such as latency, complexity, throughput, etc. Once a suitable architecture is selected, implementation is performed leading to a prototype that can be demonstrated showing the benefits of the developed idea. The hardware architectures and the demonstrator can be the subject of a second dissemination step..

(23) 4. Curriculum Vitæ. A third phase can take place where the know-how is transfered to an industrial partner to go one step further towards a commercial product. This widens the reach of these benefits to the general public.. 4. Advisorship/Tutorship. Figure 3: Summary of advisorship/tutorship activity Since my PhD defense early 2008, I have been involved in the tutorship of 9 defended PhDs, 4 Master thesis students and 8 undergraduate student projects. I am currently the advisor of 5 PhDs with a scheduled defense date ranging between the end of 2020 and the end of 2022. In parallel, I have also collaborated closely with 7 post-doctoral fellows and 4 R&D engineers. The tutorship activity is summarized in Fig. 3 and falls within the 4 main research topics shown in the legend, namely:.

(24) 5. 5. Contribution to research projects. •. Interaction between algorithm/hardware architecture.. •. Algorithms for coded modulations with or without MIMO.. •. Algorithms for filtered waveforms.. •. Algorithms for Non-Orthogonal Multiple Access (NOMA) systems.. 5. Contribution to research projects. Since 2006, I have actively participated to numerous R&D projects as summarized in Table 1. Several projects received distinction and/or awards thanks to their impact on standardization and/or thanks to their innovative aspect. Table 1: List of research projects. Type of project French national ANR Industrial with Newtec EU H2020 beyond 5G Industrial with Newtec Industrial with Orange Labs and TCL PHC CEDRE Campus France Patent factory with France brevets Patent factory with France brevets Industrial with Orange Labs EU H2020 5G-PPP. EU FP7 FUI GREENCoMM. Title Quasi-Cyclic Small Packet (QCSP) Design of improved turbo decoders Enabling Practical Wireless Tb/s Communications with Next Generation Channel Coding (EPIC) Design of improved turbo codes Implementation of a VLIW hardware accelerated 5G LDPC decoder Green and robust digital communications Patent factory for non orthogonal multiple access schemes Patent factory for filter bank multi-carrier modulation and demodulation Improving turbo codes for future standards Flexible air interface for scalable service delivery within wireless communication networks of the 5th Generation (FANTASTIC-5G) Mobile and wireless communications Enablers for Twenty-twenty (2020) Information Society (METIS) GREEN Computing and CoMMunications. Role within IMT Atlantique. Budget Period. Technical manager. 75ke. 2019-2022. Technical and admin. manager. 30ke. 2019. Setting up, WP leader and technical contribution. 416ke. 2017-2020. 30ke. 2018. 80ke. 2017-2018. 18ke. 2017-2018. Technical admin.. 300ke. 2016-2019. Technical admin.. 150ke. 2015-2018. Technical and admin. manager. 180ke. 2015-2018. Setup, technical contribution to several WPs. 486ke. 2015-2017. Setup, technical contribution to several WPs. 578ke. 2012-2015. Technical contribution. 280ke. 2012-2015. Technical and admin. manager Admin. manager and technical contribution Setup, technical contribution.

(25) 6. Curriculum Vitæ. Industrial with Orange Labs Bilateral with French space agency CNES Bilateral with French space agency CNES Industrial with SIMPULSE (SME). Improved end-to-end 3G/4G simulator. Technical manager. 25ke. 2014. Rotated constellations for satellite to mobile communications. Technical and admin. manager. 84ke. 2012-2013. Development of a EMBMS simulation chain. Technical and admin. manager. 32ke. 2011. Hardware implementation of a DVB-T2 receiver. Technical contribution. 30ke. 2011-2012. EU Eurostars. Enhanced wireless Technologies for News and Security Applications (ETNA). 131ke. 2011- 2013. French national ANR. Mobile MultiMedia (M3). Setup, technical contribution to several WPs Setup, technical contribution to several WPs. 289ke. 2011- 2013. 325ke. 2010- 2012. 174ke. 2008- 2011. 286ke. 2007- 2009. 80ke. 2007- 2008. 30ke. 2006- 2007. Celtic Plus EU. EU Eurostars Celtic EU EU space agency ESA Internal IMT. 6. Enabling Next GeneratIon NEtworks for broadcast Services (ENGINES). Received Celtic bronze award of excellence in 2014. Small and Medium Enterprise for T2 (SME42). Ranked #1 among 113 accepted projects. Broadcast for the 21st Century (B21C). Received Celtic silver award of excellence in 2009. Broadband Satellite Digital Transmissions Ultra Wide Band for Domotics (UWB-DOMO). Setup, technical contribution to several WPs Setup, technical contribution to several WPs Setup, technical contribution to several WPs Technical contribution Technical contribution. Collaborations. During my professional career, I have been able to build 3 types of collaborations: •. Strategic agreements/partnerships: recurrent collaborations within bilateral and collaborative projects that can lead to joint patent filing(s) and/or joint contributions to standards.. •. Industrial collaborations: collaborations with industrial partners in the context of research projects or a joint vision regarding a particular technique in the context of standardization.. •. Academic collaborations: repeated collaborations with academic partners leading to a strengthened relationship with common publications.. These collaborations span several countries as shown in Fig. 4. The corresponding details are as follows:.

(26) 7. 6. Collaborations. Figure 4: Collaboration map •. Strategic agreements/partnerships: • Orange Labs: Agreement on the design of enhanced turbo codes for next generation standards. • France Brevets: Agreement on financing research activity related to waveform and NOMA.. •. Industrial collaborations: • Orange Labs: collaborations within more than 8 EU and bilateral projects. • TEAMCAST (Now ENENSYS group): collaboration within 3 EU and 1 French national projects. • Newtec: collaboration within 1 European Space Agency (ESA) project EU, 2 bilateral projects and one Cifre PhD. • Panasonic Germany: collaboration during DVB-T2 and DVB-NGH standardization. • European Space Agency (ESA): collaboration during DVB-RCS2 standardization and one bilateral project.. •. Academic collaborations: • Lebanese university: Collaboration with Prof. Joumana Farah (associate researcher with IMT Atlantique): 4 common PhD students (2 defended). Execution of a joint CEDRE project (Campus France) and 11 common publications. • Holy Spirit University of Kaslik (USEK) Lebanon: collaboration with Associate professors Marie-Rita Hojeij and Charles Yaacoub: 1 PhD student and 2 Master students in common. • Basque Country University in Spain UPV/EHU: collaboration with Professor Pablo Angueira Buceta: 3 months visit of PhD student Jon Barrueco to Brest. 2 common publications..

(27) 8. Curriculum Vitæ. • Technical University of Kaiserslautern (TUK) in Germany: collaboration with professor Norbert Wehn: 1 month visit of my PhD student Ahmed Abdmouleh to TUK. Two visits of two weeks each of postdoc Stefan Weithoffer to IMT Atlantique. Stefan participated to IMT teachings. 1 common publication with 2 planned submissions. Stefan was lately hired as an associate professor at IMT Atlantique. • ENSTA Paristech: collaboration with Professor Benoît Geller and postdoc Jianxiao Yang: visit of 2 weeks of Jianxiao to IMT Atlantique. 3 common publications. • Ecole Technique Supérieure and Polytech. Montréal: collaborations with Professor Georges Kaddoum and associate professor François Leduc-Primeau (associate researcher at IMT Atlantique) and visit of François for 2 weeks to IMT Atlantique. 2 common publications. • University of Southampton: collaboration with Professor Rob Maunder: Multiple joint contributions to 3GPP technical group regarding coding for 5G. • Concordia University: collaboration with Professor Walaa Hamouda: Visit of my PhD student Ammar el Falou to Québec for 3 months. 3 common publications. • University of Illinois at Urbana-Champaign: Collaboration with Professor Venugopal V. Veeravalli: Visit of my PhD student Marie-Josepha Youssef to UrbanaChampaign for 6 months. 1 common publication and several under preparation.. 7. Remarkable results. •. In 2007, proposal of a coded Continuous Phase Modulation (CPM) scheme particularly suited for satellite transmissions. The proposal was implemented in Sat3Play modems of the company Newtec. This series of modems was successfully commercialized by the satellite operator SES Astra since 2009 using the service ASTRA2Connect.. •. Adoption of 16-state double-binary turbo codes in 2009 (for the return channel), in 2012 (for the download channel) and of a variant of the coded CPM in the DVB-RCS2 standard.. •. Delivered patent on rotated constellations technique in February 2009. Extension to Europe and Asia in 2010.. •. Adoption of rotated constellations technique and acknowledgment of the essentiality of the patent to the DVB-T2 standard in June 2008 and to its mobile version in February 2011. This adoption was extended to include the hybrid terrestrial-satellite standard for mobile broadcasting DVB-NGH in September 2011.. •. Implementation of the first published hardware prototype for a DVB-T2 receiver including rotated constellations. It was demonstrated at the International Broadcasting Convention (IBC) in 2010.. •. The publication detailing an improved DVB-T2 receiver received the best PhD paper award at the IEEE International Workshop on Signal Processing Systems (SIPS) conference in 2011.. •. Implementation of the first hardware prototype for a filtered multi-carrier waveform. The corresponding demonstrator was selected by the METIS project as one of its two contributions to the Mobile World Congress (MWC) in 2015. The corresponding booth received more than 1000 visitors including the European commissioner in charge of.

(28) 9. 8. Teaching activities. the digital technologies and the director of research and strategy at Orange Labs. A following demonstration took place at the research forum of Orange Labs. •. In 2015-2016, agreement over 3 patent factories (2 with France brevets and 1 with Orange Labs) spanning 3 different technical topics: Filtered waveforms, NOMA and coding. These topics were considered of high importance for 5G and future communications standards.. •. The publication addressing the association of NOMA techniques with single-user MIMO received the best paper award at the ISCC conference in 2017.. •. The 2019 URSI Radioscience PhD Thesis Prize was awarded to Jérémy Nadal for his work entitled: “Filtered multicarrier waveforms in the context of 5G: novel algorithms and architecture optimizations”. The PhD thesis was carried out within the IAS team under the direction of Amer Baghdadi and my supervision. The “Radioscience PhD Thesis Prize” is intended to reward annually excellent doctoral work in one or more of the URSI’s scientific fields.. 8. Teaching activities. 8.1. Topics and courses. My teaching activities took place solely at Telecom Bretagne/IMT Atlantique premises with a total average volume of around 145 hours of equivalent tutorial classes. They are related to the following topics: •. Digital electronics.. •. Components and architecture of transmission systems.. •. System on chip.. •. X86 processor architecture.. •. Error correcting codes and coded modulations.. The following table details the type of courses, their level and the volume in number of hours annually:. Table 2: Detailed teaching activities. Course. Level. Volume (eq. tut). Basic Electronics ELP203, ELP314, ELP213. 1st and 2nd year. 55 h/y. Architecture of transmission systems F14B201. 3rd year, Master. 27 h/y. ... continued in the next page.

(29) 10. Curriculum Vitæ. Table 2 – Follow-up of the previous page. Course. Level. Volume (eq. tut). Systems on chip. 3rd year. 18 h/y. Embedded systems. 3rd year, Master. 15 h/y. Introduction to engineering systems. 1st and 2nd year. 12 h/y. Discovering electronics project. 1st year. 18 h/y. 8.2. Teaching and administrative responsabilities. a) I was attributed the following responsabilities at Telecom Bretagne/IMT Atlantique: • Head of the digital and analog integrated circuit course since 2019 (40 face-to-face hours): Modification of the exercices and adaptation of the contents to the new cursus. This module takes place twice a year. • Head of the basic Electronics course ELP203 from 2014 till 2019 (21 face-toface hours): Modification of the exercices and the creation of a new course on memories. This module took place twice a year. • Main contact for continuous courses at the Electronics department since 2012. • In charge of the coding and modulation courses and exercices for 3rd year and Master degree students since 2015. b) I have also managed student research projects of 1st and 3rd year/Master degree (≈2 projects/year in average).. 8.3. Other responsabilities, activities and distinctions. a) Active participation to the code selection group in 3GPP for 5G. More than 5 technical contributions in 5 different meetings. b) Senior IEEE member since August 2019. c) Member of the technical program committee of: • The International Symposium on Turbo Codes (ISTC) initially planned in 2020 then postponed to 2021 due to Covid-19. • The IEEE Symposium on Computers and Communications (ISCC) in 2020. • ISTC in 2018..

(30) 11. 9. Publications. • ISCC in 2017. • Colloque francophone de traitement du signal et des images (GRETSI) 2017. • ISTC in 2016. d) Co-organiser of special sessions at: • ISTC in 2018 on ultra-high throughput decoding. • International Conference on Telecommunications (ICT) in 2018 on NOMA techniques. e) Chairman of the following sessions: • On the construction of LDPC codes at ISTC in 2018. • On receiver design and sensor networks at the Workshop on Signal Processing Systems (SiPS) in 2011. f) Regular reviewer for IEEE transactions on wireless communications, IEEE transactions on communications, IEEE transactions on vehicular technology, IEEE access, IEEE transactions on broadcasting, IEEE communications letters. Int. Conference on Telecommunications ICT, IEEE GLObal TeleCOMmunications conference GLOBECOM, Workshop on Signal Processing Systems SiPS, IEEE International Conference on Communications ICC. g) Promoter of scientific progress and development through the participation to the researcher’s night in Brest for several years.. 9. Publications. Figure 5: Google scholar citations and h-index The results of my research activities generated 18 patent filings and were published in 24 journal publications, 2 book chapters, 65 conferences, 2 invited talks and 15 demonstration/exhibition/talk events. The performed work received 1172 citations with a h-index of 17 (ref. Google scholar, 08/09/2020) as shown in Fig. 5. In addition, I am also a co-author of 3 standardization documents: •. The ETSI specification document EN 302 755 of the DVB-T2 standard.. •. The ETSI implementation guidelines document TS 102 831 for the DVB-T2 standard.. •. The ETSI specification document EN 303 105 for the DVB-NGH standard..

(31)

(32) Introduction to my research activities The research activities since the beginning of my PhD have addressed the following 4 main topics as can be seen in Fig. 6: •. Algorithms for coded modulations with or without Multiple-Input Multiple-Output (MIMO) techniques. This activity represents the largest part of my research. It was the subject of 6 PhDs and one post-doctoral work. It spans the aspects of code design for binary and non-binary codes, the derivation of simplified decoding algorithms, the association of coding schemes with high order modulations, the increase of the diversity degree for coded modulations over fading and erasure channels, the association of coding schemes with multi-antenna systems and the design of coding/decoding schemes for high throughput applications.. •. Algorithms for filtered waveforms. An important part of my research activity targets the study of post-OFDM filtered waveforms. This topic spans the design of such waveforms at the transmitter and the receiver side while addressing their penalizing drawbacks. This work was the subject of two PhDs and two post-doctoral fellowships.. •. Algorithms for NOMA systems. This represents the newest topic of my research activity. It represents a shift in the type of treated topics addressing mainly the physical layer of communications and broadcasting systems to higher layers. Indeed, by studying the scheduling and the derivation of resource and power allocation schemes for complex systems with orthogonal and non-orthogonal multiple access, multiple antennas (centralized or distributed) with or without coordinated joint transmissions, we address MAC and crosslayer problems. This work is performed by 3 PhDs and is carried out in collaboration with Prof. Joumana Farah from the Lebanese university.. •. Interaction between algorithm/hardware architecture. At the exception of resource allocation-based algorithms generally applied via software on general purpose processors, my research activity in the algorithmic domain is performed while taking into account potential future implementations. Therefore, it is natural to dedicate a considerable part of my research activity to the implementation of the derived algorithms. Indeed, in collaboration with colleagues specialized in hardware implementations, 4 PhDs took place on following the IAS approach on algorithms addressing coded modulations and waveform design.. In the following, for each topic a motivation for the performed work will be given, showing what is missing from the state-of-the-art, followed by the list of contributions through the performed technical work. Most important results are shown next followed by one or 13.

(33) 14. Introduction to my research activities. Figure 6: Map of the research activities since the beginning of my PhD more impacting publications on the subject. Finally, the dissemination effort regarding the obtained results is provided. Performed work on implementation targets algorithms proposed for one of the other topics as shown in Fig. 6. Therefore, the implementation work will be presented in the codingrelated and waveform-related sections..

(34) Chapter. Algorithms for coded modulations with(out) MIMO. 1.1. Introduction and motivation. After my PhD defense in 2008, I have been deeply involved in the definition of Ffour different second-generation Digital Video Broadcasting (DVB) standards: the terrestrial fixed and mobile components DVB-T2 [1], DVB-T2 Lite [1] and DVB-NGH [2], in addition to the second generation of Return Channel Satellite DVB-RCS2 [3]. The work on the terrestrial fixed and mobile components was a direct extension of the second part of my PhD work, which focused on coded modulation schemes applying advanced constellation design and signal space diversity, or namely “rotated constellations”, to approach capacity and improve robustness against severe channel conditions with deep fades. Through my extensive involvement in several Celtic EU (B21C and ENGINES), EU Eurostars (SME42 and ETNA), ANR (M3) projects and collaborations with other academic partners ranging from late 2007 to 2013 (until late 2017 for the collaborations) as detailed in Table 1, I had the chance to personally work on the refinement of the work done during my PhD and to co-supervise several PhD students and post-doctoral fellows as shown in Fig. 6. To further increase throughput of real systems, the study of the introduction of multiple antenna schemes in the presence of channel coding complemented these studies at a later stage. After the definition of second generation broadcasting standards, the focus shifted to the definition of 5G systems and beyond. Of a different type, the study items were naturally different from the ones for broadcasting standards with different sets of constraints. In concrete terms, the accent was put simultaneously on designing flexible (in frame size and coding rate) coding schemes, on achieving low error floors and on attaining high throughput with efficient hardware. In the context of the PhDs of Ronald, Rami, Vinh Hoang Son and Titouan, we have tried to achieve these multiple objectives by focusing on the design of binary and non-binary Turbo codes. Note that the non-binary variant may also be suitable for next generation broadcasting systems due to the increased capacity of symbol-based coded modulations when compared to their bit-based counterpart and to their improved asymptotic performance. 15. 1.

(35) 16. 1.2 1.2.1. Algorithms for coded modulations with(out) MIMO. Diversity, constellation and mapping for broadcasting standards Context and prior art. The emergence of new market driven services such as high definition television and 3DTV have offered unprecedented user experience creating a real need for improving transmission systems. A better use of the scarce spectrum resources became a must, leading to the development of the second generation broadcasting systems. Single Frequency Network (SFN) is a way to increase spectral efficiency. It consists of a broadcast network where several transmitters simultaneously send the same signal over the same frequency channel. While spectrally efficient, such a topology can lead to a severe form of multipath propagation. Indeed, the receiver sees several echoes of the same signal and the destructive interference among these echoes known as self-interference may result in additional fade events. This is problematic especially in wideband and high data rate digital communications, since the frequency-selective fading and the Inter-symbol Interference (ISI) caused by the time spreading of the echoes greatly deteriorate system performance in terms of Bit Error Rate (BER). Spectral efficiency should not come at the price of reduced robustness. Therefore, numerous technical aspects are to be improved from first generation systems including source coding, channel coding, interleaving, modulation, diversity etc. In 2008, the work on the definition of the DVB-T2 standard was launched. This second generation introduced several enhancements to the transmission system including the 4th generation of the Moving Picture Experts Group (MPEG4) source coding, multiple physical layer pipes, and a stateof-the-art forward error correcting (FEC) code: Low Density Parity Check (LDPC) [4] + Bose Ray-Chaudhuri Hocquenghem (BCH) [5]. Incresed diversity was also introduced by a longer channel interleaver and the adoption of a diversity technique at the signal space level, a Multiple Input Single Output (MISO) Alamouti [6] based-scheme, etc. The contributions related to this context are described in section 1.2.2, but as a first step, the following sub-sections describe the prior state of the art and/or the issues to be solved for each topic under consideration.. 1.2.1.1. Coded modulation and diversity techniques. The work performed in this context focuses on signal-space diversity techniques and their association with the FEC decoder at the receiver side. Reliable transmissions over channels with deep fades require the increase of the diversity order. In 2008, the Bit-Interleaved Coded Modulation (BICM) principle [7] introduced by Zehavi in [8] represented the state of the art in coded modulations over fading channels. At the receiver side, an iterative exchange of information between the FEC decoder and the constellation demapper can be implemented in order to further improve the reliability of the transmission. This was first studied in [9] where a BICM scheme with Iterative Demodulation or Demapping (BICM-ID) was proposed, using a convolutional code with additional soft feedback from the Soft-Input Soft-Output (SISO) decoder to the constellation demapper. Later, in [10], the authors investigated different mapping techniques suited for BICM-ID and 16-QAM constellations. They proposed several mapping schemes providing large coding gains. In [11], the convolutional code classically used in BICM-ID schemes was replaced by a Turbo code, but only a small gain of 0.1 dB was observed. These findings suggested that BICM-ID was not appropriate for Turbo-like coding solutions even though the added complexity is relatively small. On another note, the Signal Space Diversity (SSD) principle introduced in [12, 13] is a means of increasing the diversity order of BICM schemes over fading channels. While a known.

(36) 1.2. Diversity, constellation and mapping for broadcasting standards. 17. concept since 1997, this principle had never been studied and published with advanced FEC code solutions such as Turbo or LDPC codes before our work [14]. 1.2.1.2. Signal space diversity and receiver implementation. Thanks to our work and action in standardisation, SSD or more precisely rotated constellations were adopted in the DVB-T2 standard in 2011. Logically, the need for efficient hardware implementations followed. The main objective of a study we initiated on the topic was to design a DVB-T2 compliant receiver with an acceptable hardware complexity. While improving performance, SSD introduces additional complexity especially for spectrally efficient constellation sizes. DVB-T2 was the first standard to adopt signal space diversity with high order constellations such as 256-QAM. In this case, the classical one dimensional Max-Log demapping algorithm applied on log(M ) Pulse Amplitude Modulation (PAM) based on de-coupling the I and Q components is not applicable. The quest for a hardware efficient SSD demapper was raised and not addressed before our study. On the other hand, the implementation of the LDPC codes adopted in DVB-T2 is tricky, due to the double diagonal sub-matrices observed in their parity-check matrices. They induce memory update conflicts in the shuffled LDPC decoding architecture, which consequently causes inefficient message passing between the check nodes and bit nodes. These are crucial problems that have to be addressed when designing an LDPC decoder dedicated to the DVBT2 standard. A classical iterative receiver is frame-based, which induces large latency. The latency is introduced by the interleaving/de-interleaving block, which is based on memory writing and reading. The latency is also due to the state-of-the-art LDPC decoding algorithm (horizontal layered decoding algorithm). Indeed this algorithm provides the extrinsic information only after one complete iteration. Therefore, one iteration of a classical receiver consists of one complete iteration of LDPC decoding, de-interleaving memory writing and reading, demapping and interleaving memory writing and reading. The resulting large latency prohibits any efficient message exchange between the demapper and decoder, hence reducing the throughput. All these issues were addressed in the context of the PhD thesis of Meng Li supported by the EU Eurostars SME-42 project. The designed receiver supports both non-iterative process and iterative process. The iterative receiver was studied even though practical applications are generally reluctant to mandate solutions based on iterative processes due to some challenges and constraints in terms of increased hardware complexity, memory access conflicts and additional latency. 1.2.1.3. Channel emulation/simulation. Achieved gains by SSD greatly depend upon the encountered channel. Therefore, special attention was put on the accurate simulation of correlated fading channels. State-of-the-art Rayleigh fading simulators can be divided into three main categories: the Sum of Sinusoids (SOS) method, the filtering white Gaussian variables method and the Inverse Fast Fourier Transform (IFFT) method. The first method described in [15, 16] involves the superposition of a number of sinusoidal wave components. Each component is represented by a group of amplitude, frequency, and phase values related to the Doppler spectrum. Once the group of values is initialized at the beginning of each simulation trial, it is then kept constant during the whole duration of the trial. When an on-the-fly sample generation is required, the number of independent sinusoid functions needed to generate statistically accurate Doppler variables is too high to be supported by real-time implementations. In addition, the periodic nature of the underlying.

(37) 18. Algorithms for coded modulations with(out) MIMO. sinusoidal function introduces an unexpected correlation between samples when a long interval of time has to be simulated. In order to reduce the number of sinusoidal components, a low-complexity channel simulator was proposed in [17] by using a truncated subspace representation of the Doppler spectrum based on discrete prolate spheroidal sequences. This technique is useful for a hardware emulator, but only if a limited numerical precision is required. The second approach consists of filtering a white Gaussian sequence by using a Doppler filter in time domain. It can be divided into two different methods depending on the filter types. In [18], a Finite Impulse Response (FIR) filter is used whereas in [19], an Infinite Impulse Response (IIR) filter is proposed. FIR filter-based simulators do not suffer from any numerical instability. However, the number of Doppler filter taps required to meet statistical requirements leads to a high computational complexity when the ratio between the sampling frequency and Doppler frequency is high. Whereas the IIR filter-based simulators could provide a solution with lower complexity, they do suffer from numerical instabilities. Nevertheless, some solutions are provided in [19] in order to mitigate this risk. The third approach, based on the IFFT method, is described in [20, 21]. The corresponding simulator first multiplies a series of independent complex Gaussian variables by a frequency mask corresponding to the square root of the Doppler spectrum. Then, it performs an IFFT over the frequency domain sequence and obtains the time domain Doppler variates. Thanks to the use of IFFT, this approach is the most computationally efficient of all previous methods while providing remarkable statistical properties. Unfortunately, one important drawback resides in its block-oriented nature requiring all Doppler variates to be generated by a single IFFT and stored in advance. Memory requirements exclude its application to continuous transmissions and high sampling frequency systems with long simulation periods. To summarize this part, an efficient approach for simulating correlated fading was still lacking in literature. Here is why we decided to tackle this issue. 1.2.1.4. Non-uniform constellations and corresponding demappers. A collaboration with professor Pablo Angueira from the University of the Basque Country (UPV/EHU) was initiated in the context of the PhD of Jon Barrueco on the possible improvements of a DVB-T2-like system for the definition of the third generation US broadcasting standard, ATSC 3.0. In this new generation standard, Non-Uniform Constellations (NUCs) [22] were introduced to provide performance gains up to 1.8 dB compared to classical QAMs. NUCs are designed to maximize the BICM channel capacity [23, 24, 25]. There are two possible NUC families: 1-Dimensional NUC (1D-NUC) and 2-Dimensional NUC (2D-NUC). 1D-NUCs were already included in DVB-NGH [26]. For uniform constellations, the standard algorithms for searching the closest point to the received observation can usually be simplified for easy hardware implementation. However, for NUCs, these simplified algorithms cannot be applied directly, increasing demapping p complexity. Most existing demappers resort to exhaustive search with a complexity in O( (M )) where M is the order of the constellation. Unlike 1D-NUCs, 2D-NUCs cannot be created using two non-uniform PAM signals. This implies that the in-phase (I) and quadrature (Q) components of the constellation symbols are no longer independent. As a consequence, a two-dimensional demapper (2D-demapper) must be applied to each received symbol. In addition, the number of Euclidean distances to be computed for Maximum Likelihood (ML) demapping of 2D-NUCs increases notably with respect to the one required for demapping 1D-NUCs [27]. This fact results into a significant increase in demapping complexity for 2D-NUCs when compared to 1D-NUCs. However, 2DNUCs provide larger spectral efficiency gains, approaching closely Shannon channel capacity limit for higher constellation orders. Therefore, ATSC 3.0 includes 2D-NUCs up to 256.

(38) 1.2. Diversity, constellation and mapping for broadcasting standards. 19. constellation points. Nevertheless, higher order 2D-NUCs (1K, 4K) were discarded during the standardization process due to their excessive hardware complexity. A literature review showed abundant proposals for complexity reduction in demapping algorithms for MIMO [28, 29, 30] and rotated constellations [31, 32, 33, 34]. However, prior to our work, there was only one technical contribution addressing demapping techniques for 2D-NUCs [27]. Now adopted in a standard, simplified high order 1D and 2D-NUC demappers became a practical need. 1.2.1.5. Advanced interleaving for improved system performance. State-of-the-art FEC codes are now able to approach Shannon capacity [35, 36], especially for transmissions over memoryless channels where errors are randomly distributed and statistically independent. In contrast, when FEC codes are used in transmissions over channels in which the signal undergoes impulsive noise (i.e., error bursts in a short period of time) and/or selective fading (i.e., interference in a short frequency interval), the error rate performance can be greatly degraded. One possible solution to such a problem involves spreading error patterns arriving in bursts among several FEC frames. Therefore, channel interleaving [37] can be introduced to uniformly distribute codewords in time and frequency, in such a way that the transmitted symbols subject to impulsive noise and selective fading do not end up in the same coded frame. In other words, channel interleaving allows the overall system to take advantage of the available time and frequency diversities from the encountered channel. Although DVB-T included the OFDM modulation technique, the lack of time interleaving made the broadcast system perform poorly in mobile environments [38]. To overcome this limitation, the next generation DVB-T2 includes more elaborate channel interleaving structures. Indeed, DVB-T2 advocates the use of three interleavers [1]: cell interleaver, time interleaver and frequency interleaver. The implementation of each of them introduces further system latency. On the other hand, since the implementation of each type of interleaver requires a specific part of the memory, the whole interleaving memory can represent a large part of the silicon area and of the power consumption of the receiver. In addition, from a diversity point of view, a bad interaction between channel interleavers can introduce a degradation of the overall system performance. To deal with these issues, a new technique to analyze and design channel interleavers was developed.. 1.2.2 1.2.2.1. Performed work and contributions Signal space diversity design and action in DVB standardization. The performed work in the context of the DVB-T2 standard addressed first the issue of properly designing and associating SSD with a powerful outer FEC codes in order to achieve a level of gains justifying the paid corresponding demapper complexity. This was done during my PhD and my postdoctoral fellowship. It focused on the choice of the design criteria that were used to define the rotation angle of the QAM contellation. Moreover, an iterative process between a FEC decoder and a soft MIMO detector [39] or a demapper or an interference canceller has proven to improve performance. The iterative process between a rotated QAM demapper and a LDPC decoder was shown to improve performance after a careful selection of the iterative process scheduling. A description of the work, the considered system model and studied channel models are summarized in the publication entitled “Improving BICM Performance of QAM constellations for broadcasting applications” included hereafter and published at the International Symposium on Turbo Codes and related topics (ISTC) in 2008, which received more than 60 citations in Google scholar. An extension of the results was also published in [40]..

(39) 20. Algorithms for coded modulations with(out) MIMO. DVBSCENE The Technology Update from DVB . Issue No.. The work performed in the included publication was the subject of several technical contributions to the DVB-T2 standard. A specific working group led by the British Broadcasting Corporation (BBC) was created specifically to finalize the technical aspects related to the adoption of rotated constellations and the DVB-T2 recommendations. The choices made in the publication have shown gains reaching more than 5.0 dBs under some severe cases. Therefore, they were adopted in the standard and the iterative process at the receiver side was recommended in their implementation guidelines in order to improve the performance over fading channel without and with erasures. The fading channel with erasures represents the case of a severe fading in SFN networks. These choices were also the subject of a patent filing in 2008 [41] that was delivered and recognized as essential for the standard. This adoption brought a wide spread recognition, indeed a simple Google search of both terms Télécom Bretagne and DVB-T2 results in more than 1500 technical and news articles online. We can say that rotated constellations became the destinguishing technique used to promote the DVB-T2 standard as attested by the DVB scene magazine cover from 28/02/2011 shown in Fig. 1.1.. 37. March 2011 www.dvb.org. ”DRAMATICALLY BETTER THAN ANY OTHER DIGITAL TERRESTRIAL TRANSMISSION SYSTEM” ”DATA RATES THAT ARE 50% GREATER”. ”PUSHING THE BOUNDARIES OF TECHNOLOGY”. ”REMARKABLE SUCCESS STORY”. GLOBAL SMASH HIT STARRING:. ROTATED CONSTELLATIONS • MULTIPLE PHYSICAL LAYER PIPES • ALAMOUTI CODING EXTENDED INTERLEAVING • FUTURE EXTENSION FRAMES Semiconductor solutions for 2nd generation DVB receivers. The Road to DVB-T2 in Southern Africa. 05. DVB-T2 in Finland. 08. 11. 04 2 nd Generation DVB Interactive Satellite System 06 DVB-T2 Hybrid Solution 12 DVB-3DTV Specification Approval 13 A nalysis 14 Market Watch. Figure 1.1: DVB Scene magazine cover from 28/02/2011 with emphasis on rotated constellations..

(40) 1.2. Diversity, constellation and mapping for broadcasting standards. Improving BICM Performance of QAM constellations for broadcasting applications Charbel Abdel Nour and Catherine Douillard TELECOM Bretagne (Institut TELECOM), Lab-STICC Laboratory Brest, France Email: charbel.abdelnour@telecom-bretagne.eu and catherine.douillard@telecom-bretagne.eu. Abstract— A technique intended to improve the performance of bit-interleaved coded modulations over non Gaussian channels is presented. It introduces signal space diversity by the means of modifications to the constellation mapper and to the corresponding demapper. A rotation of the constellation is followed by a signal space component interleaving. Iterative processing at the receiver side is shown to provide additional error correction. This method outperforms state-of-the-art systems over flat fading channels and substantial gains with respect to bit-interleaved coded modulation are obtained for severe channel conditions. It has been adopted by the Digital Video Broadcasting European consortium in the upcoming next generation of digital terrestrial television, DVB-T2, for the 4-, 16-, 64- and 256-QAM constellations. The resulting improvement in performance can vary from 0.2 dB to several dBs depending on the order of the constellation, the coding rate and the channel model.. I. INTRODUCTION Next generation broadcast systems should be designed to satisfy the need of high data rate transmissions through improving the robustness to severe channel conditions. Reliable transmissions over channels with deep fades require the increase of the diversity order. The Bit-Interleaved Coded Modulation (BICM) principle [1] introduced by Zehavi in [2] currently represents the state-of-the-art in coded modulations over fading channels. The Bit-Interleaved Coded Modulation with Iterative Demodulation or Demapping (BICM-ID) scheme proposed in [3] is based on BICM using a convolutional code with additional soft feedback from the Soft-Input Soft-Output (SISO) decoder to the constellation demapper. In [4], the authors investigated different mapping techniques suited for BICM-ID and 16-QAM constellations. They proposed several mapping schemes providing large coding gains. In [5], the convolutional code classically used in BICM-ID schemes was replaced by a turbo code. Only a small gain of 0.1 dB was observed. This result makes BICM-ID with turbo-like coding solutions unsatisfactory even though the added complexity is relatively small. The Signal Space Diversity (SSD) principle introduced in [67] improves the diversity order of BICM schemes over fading channels. While a known concept since 1997, this principle This work has been carried out in the framework of the CELTIC B21C European project.. had never been studied and published with advanced Forward Error Correcting (FEC) code solutions such as turbo or LDPC codes before [8]. In this paper, the authors investigated the impact of introducing the SSD principle with a 16-QAM constellation and the Digital Video Broadcasting Return Channel Satellite (DVB-RCS) turbo code over a flat fading Rayleigh channel. They have shown by means of Extrinsic Information Transfer (EXIT) charts [9] that an additional improvement in performance is achievable when iterative demapping is applied even in the case of using turbo codes. In our work, modifications to the original proposal of [6-7] are made in order to adapt the principle to the outer code solution and to increase the robustness of the receiver to severe channel conditions. For this purpose, deep fades have been introduced into the channel model, that can be seen as erasure events at the receiver side. In addition, another light is shed onto iterative demapping when associated with the SSD principle explaining the additional improvement in performance. This paper is organized as follows: In section 2, we present the channel model considered in this study. In section 3, we start by a brief description of the BICM principle followed by a presentation of the studied SSD method. Introduced modifications to the mapper and its associated demapper are then detailed. We finally show that the proposed SSD technique can be viewed as the application of a repetition code, thus justifying the interest of iterative demapping at the receiver side. Section 4 describes the adaptation of this SSD technique to actual transmissions systems, especially for broadcasting applications. In particular, it details the criteria adopted for the choice of the rotation angle. Section 5 presents some simulation results corresponding to DVB-T2 transmission scenarios. Section 6 concludes the paper. II. THE CHANNEL MODEL Two different channel types have been considered: the classical Rayleigh fading channel and a variation of this channel to which we have added additional deep fade events. Both are based on the fading channel model of [8]. The channel model with deep fades is representative for broadcasting applications over Single Frequency Networks. IEEE Symposium on turbo codes and iterative processing, 2008. 21.